I
ndo
ne
s
i
a
n J
o
ur
na
l
o
f
E
l
e
c
t
r
i
c
a
l
E
ng
i
ne
e
r
i
ng
a
nd C
o
m
p
u
t
e
r
Sc
i
e
nc
e
V
o
l.
11
,
N
o.
1
,
J
ul
y
201
8
, p
p
.
294
~
29
9
I
SSN
:
2502
-
4752
,
D
O
I
:
10.
11
591/
i
j
eecs
.
v
11
.i
1
.p
p
294
-
2
99
294
Jou
r
n
al
h
om
e
p
age
:
h
ttp
:
//ia
e
s
c
or
e
.
c
om
/
j
our
nal
s
/
i
nde
x
.
php/
i
j
e
e
c
s
E
ns
e
m
bl
e M
et
ho
d f
o
r I
ndo
nes
ia
n
T
w
it
t
er
H
a
t
e Sp
e
ech
Det
ec
t
io
n
M
.
A
li F
a
u
z
i
1
,
A
nny
Y
uni
a
r
t
i
2
1
F
acu
l
t
y
o
f
C
o
m
p
u
t
er
S
ci
en
ce,
B
r
aw
i
j
a
y
a U
n
i
v
er
s
i
t
y
,
M
al
an
g
,
I
ndo
ne
s
i
a
2
I
nf
or
m
a
t
i
c
s
D
e
pa
r
t
m
e
nt
,
I
ns
t
i
t
ut
T
e
k
nol
og
i
S
e
p
ul
uh N
ope
m
be
r
,
S
ur
a
ba
y
a
,
I
ndone
s
i
a
A
rt
i
cl
e I
n
f
o
AB
S
T
RAC
T
A
r
tic
le
h
is
to
r
y
:
R
ecei
v
ed
27 D
e
c
,
201
7
Re
v
i
s
e
d
J
a
n 9
,
201
8
A
ccep
t
ed
J
a
n 1
6
,
2
01
8
D
u
e t
o
t
h
e
m
as
s
i
v
e i
n
cr
eas
e o
f
u
s
er
-
g
en
er
at
ed
w
eb
co
n
t
en
t
,
i
n
p
ar
t
i
cu
l
ar
o
n
s
o
ci
al
m
ed
i
a n
et
w
o
r
k
s
w
h
er
e an
y
o
n
e can
g
i
v
e a
s
t
at
e
m
en
t
f
r
eel
y
w
i
t
h
o
u
t
an
y
lim
ita
tio
n
s
,
th
e
a
m
o
u
n
t
o
f
h
a
te
f
u
l a
c
tiv
itie
s
is
a
ls
o
i
n
c
r
e
a
s
in
g
.
S
o
c
ia
l m
e
d
i
a
a
nd m
i
c
r
obl
og
g
i
ng
w
e
b s
e
r
v
i
ce
s
,
s
u
ch
as
T
w
i
t
t
er
,
al
l
o
w
i
n
g
t
o
r
ead
an
d
an
al
y
ze u
s
er
t
w
eet
s
i
n
n
ear
r
e
al
t
i
m
e.
T
w
i
t
t
er
i
s
a l
o
g
i
cal
s
o
u
r
ce
o
f
d
at
a
f
o
r
h
at
e s
p
eech
an
al
y
s
i
s
s
i
n
ce u
s
er
s
o
f
t
w
i
t
t
er
ar
e
m
o
r
e l
i
k
el
y
t
o
ex
p
r
es
s
t
h
ei
r
e
m
ot
i
ons
of
a
n e
v
e
nt
by
pos
t
i
ng
s
om
e
t
w
e
e
t
.
T
hi
s
a
na
l
y
s
i
s
can
h
el
p
f
o
r
ear
l
y
i
de
nt
i
f
i
c
a
t
i
on
of
ha
t
e
s
pe
e
c
h s
o i
t
c
a
n be
pr
e
v
e
nt
e
d t
o be
s
pr
e
a
d
w
i
de
l
y
.
T
he
m
a
n
u
a
l w
a
y
o
f
c
la
s
s
i
f
y
in
g
o
u
t h
a
te
f
u
l c
o
n
te
n
ts
in
tw
itte
r
is
c
o
s
tl
y
a
n
d
n
o
t
s
cal
ab
l
e.
T
h
er
e
f
o
r
e,
t
h
e au
t
o
m
at
i
c
w
a
y
o
f
h
at
e
s
p
eech
d
et
ect
i
o
n
i
s
n
eed
ed
t
o
b
e d
e
v
e
l
ope
d f
or
t
w
e
e
t
s
i
n I
nd
one
s
i
a
n l
a
ng
ua
g
e
.
I
n t
hi
s
s
t
u
dy
,
w
e
us
e
d
e
ns
e
m
bl
e
m
e
t
hod f
or
ha
t
e
s
pe
e
c
h de
t
e
c
t
i
on i
n I
n
do
ne
s
i
a
n l
a
ng
ua
g
e
.
W
e
e
m
pl
oy
e
d
f
i
v
e
s
t
a
nd
-
a
lo
n
e
c
la
s
s
if
ic
a
tio
n
a
lg
o
r
ith
m
s
,
in
c
lu
d
i
n
g
N
a
ïv
e
B
a
y
e
s
,
K
-
N
e
a
r
e
s
t
N
e
i
g
hbour
s
,
M
a
x
i
m
u
m
E
nt
r
opy
,
R
a
ndo
m
F
or
e
s
t
,
a
n
d S
up
por
t
V
e
c
t
or
M
a
c
hi
ne
s
,
a
nd t
w
o e
ns
e
m
bl
e
m
e
t
hods
,
ha
r
d v
ot
i
ng
a
nd s
of
t
v
ot
i
ng
,
o
n
T
w
i
t
t
er
h
at
e s
p
eech
d
at
as
et
.
T
h
e ex
p
er
i
m
en
t
r
es
u
l
t
s
s
h
o
w
ed
t
h
at
u
s
i
n
g
en
s
e
m
b
l
e
m
et
h
o
d
can
i
m
p
r
o
v
e t
h
e cl
as
s
i
f
i
cat
i
o
n
p
er
f
o
r
m
an
ce.
T
h
e b
es
t
r
es
u
l
t
i
s
ach
i
ev
ed
w
he
n us
i
ng
s
of
t
v
ot
i
ng
w
i
t
h F
1 m
e
a
s
ur
e
79.
8%
on
unba
l
a
nc
e
da
t
a
s
e
t
a
nd 84.
7%
on ba
l
a
nc
e
d
da
t
a
s
e
t
.
A
l
t
houg
h t
he
i
m
pr
ov
e
m
e
nt
i
s
not
t
r
ul
y
r
e
m
a
r
ka
bl
e
,
us
i
ng
e
ns
e
m
bl
e
m
e
t
hod c
a
n r
e
duc
e
t
he
j
e
opa
r
dy
of
ch
o
o
s
i
n
g
a p
o
o
r
cl
as
s
i
f
i
er
t
o
b
e
u
s
ed
f
o
r
d
et
ect
i
n
g
n
ew
t
w
eet
s
as
h
at
e s
p
eech
or
n
ot
.
Ke
y
wo
rd
s
:
Cl
a
s
s
i
f
i
e
r
e
n
s
e
m
b
l
e
H
at
e s
p
eech
I
nd
o
ne
s
i
a
n l
a
n
g
ua
ge
T
e
x
t c
la
s
s
if
ic
a
tio
n
T
w
it
te
r
C
opy
r
i
g
ht
©
201
8
I
ns
t
i
t
ut
e
o
f
A
d
v
anc
e
d E
ngi
ne
e
r
i
ng
an
d
Sc
i
e
nc
e
.
A
l
l
ri
g
h
t
s re
se
rv
e
d
.
Co
rre
sp
o
n
d
i
n
g
Au
t
h
o
r
:
M
.
A
li F
a
u
z
i
,
F
acu
l
t
y
o
f
C
o
m
p
u
t
er
S
ci
en
ce,
B
r
aw
i
j
a
y
a U
n
i
v
er
s
i
t
y
,
M
al
an
g
,
I
ndone
s
i
a
E
m
a
il:
m
o
ch
.
al
i
.
f
au
zi
@
u
b
.
ac.
i
d
1.
I
NT
RO
D
UCT
I
O
N
H
at
e s
p
eech
i
s
an
y
co
m
m
u
n
i
c
at
i
v
e act
s
t
h
a
t
u
s
ed
t
o
ex
p
r
es
s
h
at
r
ed
t
o
w
ar
d
s
a p
er
s
o
n
o
r
a
g
r
o
u
p
o
n
t
h
e
b
as
i
s
o
f
s
o
m
e
ch
ar
act
er
i
s
t
i
c
s
u
c
h
as
r
ace,
et
h
n
i
ci
t
y
,
g
e
n
d
er
,
s
ex
u
al
o
r
i
en
t
at
i
o
n
,
n
at
i
o
n
al
i
t
y
,
r
el
i
g
i
o
n
,
o
r
o
t
h
er
ch
ar
act
er
i
s
t
i
c [
1
]
.
D
u
e
t
o
t
h
e
m
as
s
i
v
e i
n
cr
eas
e o
f
u
s
e
r
-
g
e
n
er
at
ed
w
eb
co
n
t
e
n
t
,
i
n
p
ar
t
i
cu
l
ar
o
n
s
o
ci
al
m
ed
i
a n
et
w
o
r
k
s
w
h
er
e an
y
o
n
e can
g
i
v
e
a s
t
a
t
e
m
e
n
t
f
r
eel
y
w
i
t
h
o
u
t
an
y
l
i
m
i
t
at
i
o
n
s
,
t
h
e
a
m
o
u
n
t
o
f
h
at
e
f
u
l
act
i
v
i
t
i
e
s
i
s
a
l
s
o
i
n
cr
eas
i
n
g
.
S
o
ci
al
m
ed
i
a
t
ech
n
o
l
o
g
y
m
a
k
e
p
eo
p
l
e ab
l
e t
o
ex
p
r
es
s
t
h
ei
r
o
p
i
ni
o
n,
i
nc
l
ud
i
n
g
h
at
e
s
p
eech
,
q
u
i
ck
l
y
,
t
h
e
n
s
p
r
ead
w
i
d
el
y
a
n
d
b
eco
m
e
v
i
r
al
i
f
t
h
e t
o
p
i
cs
co
v
er
ed
ar
e
‘
i
n
t
er
es
t
i
n
g
’
.
I
t
ca
n
b
r
i
n
g
u
p
d
i
s
p
u
t
e
s
b
et
w
een
g
r
o
u
p
s
i
n
s
o
ci
et
y
.
I
n
I
n
d
o
n
es
i
a,
b
as
ed
o
n
t
h
e d
at
a o
f
N
at
i
o
n
al
P
o
l
i
ce C
r
i
m
i
n
al
I
nve
s
t
i
ga
t
i
o
n
A
ge
nc
y o
f
I
nd
o
n
es
i
a i
n
2
0
1
5
,
t
h
er
e ar
e
1
4
3
cy
b
er
cr
i
m
e
s
i
n
t
h
e f
o
r
m
o
f
h
at
e
s
p
eech
.
T
h
i
s
n
um
be
r
i
n
c
r
e
a
s
e
d t
o 199 i
n
2016.
H
ow
e
v
e
r
,
t
h
i
s
da
t
a
o
nl
y
c
ov
e
r
h
a
t
e
s
pe
e
c
h
e
s
be
i
n
g c
r
i
m
i
n
a
l
i
z
e
d a
n
d
r
ep
o
r
t
ed
t
o
t
h
e p
o
l
i
ce.
O
b
v
i
o
u
s
l
y
t
h
er
e ar
e s
t
i
l
l
m
an
y
m
o
r
e
h
at
e s
p
eech
es
t
h
at
ex
i
s
t i
n
v
a
r
io
u
s
s
o
c
ia
l
m
e
d
ia
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nd
o
ne
s
i
a
n J
E
l
e
c
E
ng
&
C
o
m
p
Sc
i
I
SSN
:
2502
-
4752
E
ns
e
m
bl
e
M
e
t
hod f
or
I
n
done
s
i
an T
w
i
t
t
e
r
H
at
e
Spe
e
c
h D
e
t
e
c
t
i
on
(
M
.
A
li F
a
u
z
i
)
295
O
n
e
o
f
th
e
p
o
p
u
la
r
s
o
c
ia
l
m
e
d
ia
in
I
n
d
o
n
e
s
ia
is
T
w
itte
r
[
2
]
.
S
o
c
ia
l m
e
d
ia
a
n
d
m
ic
r
o
b
lo
g
g
i
n
g
w
e
b
s
er
v
i
ces
,
s
u
c
h
as
T
w
i
t
t
er
,
al
l
o
w
i
n
g
t
o
r
ead
an
d
an
al
y
ze u
s
er
t
w
eet
s
i
n
n
ear
r
eal
t
i
m
e.
T
w
i
t
t
er
i
s
a l
o
g
i
ca
l
s
o
u
r
ce o
f
d
at
a f
o
r
h
at
e
s
p
eech
an
al
y
s
i
s
s
i
n
ce u
s
er
s
o
f
t
w
i
t
t
e
r
ar
e m
o
r
e l
i
k
el
y
t
o
ex
p
r
es
s
t
h
ei
r
e
m
o
t
i
o
n
s
o
f
a
n
ev
en
t
b
y
p
o
s
t
i
n
g
s
o
m
e t
w
eet
[
3
]
.
T
h
i
s
an
al
y
s
i
s
can
h
el
p
f
o
r
ear
l
y
i
d
en
t
i
f
i
cat
i
o
n
o
f
h
a
t
e s
p
eech
s
o
i
t
can
b
e
pr
e
v
e
n
t
e
d t
o be
s
pr
e
a
d w
i
de
l
y.
I
t
i
s
a
l
s
o u
s
e
f
u
l
f
or
c
on
t
e
n
t
f
ilte
r
in
g
a
n
d
e
a
r
l
y
d
e
te
c
tio
n
o
f
w
r
o
n
g
f
u
l a
c
tiv
it
ie
s
[
4
]
.
T
he
m
a
n
ua
l
w
a
y
o
f
d
et
ect
i
n
g
o
ut
ha
t
e
f
ul
t
w
eet
s
i
s
co
s
t
l
y
a
n
d
n
o
t
s
cal
ab
l
e.
T
h
er
ef
o
r
e,
t
h
e
au
t
o
m
at
i
c
w
a
y
o
f
h
at
e s
p
eec
h
d
et
ect
i
o
n
i
s
n
ee
d
ed
t
o
b
e d
ev
el
o
p
ed
f
o
r
t
w
eet
s
i
n
I
n
d
o
n
es
i
an
l
a
n
g
u
a
g
e.
S
o
me
p
r
e
v
i
ous
w
or
k
s
pr
opos
e
d i
n
h
a
t
e
s
pe
e
c
h
de
t
e
c
t
i
on
m
os
t
l
y
f
or
E
n
g
l
i
s
h
[
5
-
7]
.
M
os
t
of
t
h
e
m
u
s
e
d
m
ach
i
n
e l
ear
n
i
n
g
t
ech
n
i
q
u
e an
d
t
h
e d
at
as
et
i
s
f
r
o
m
T
w
i
t
t
er
.
Mean
w
h
i
l
e,
t
h
e s
t
u
d
y
o
f
h
a
t
e s
p
eech
d
et
ect
i
o
n
i
n
I
n
d
o
n
es
i
a
n
l
an
g
u
ag
e
i
s
s
t
i
l
l
v
er
y
r
ar
e.
A
s
f
ar
as
w
e
k
n
o
w
,
[
8
]
an
d
[
9
]
ar
e t
h
e o
n
l
y
w
o
r
k
s
i
n
h
at
e
s
p
eech
de
t
e
c
t
i
on
i
n
I
n
don
e
s
i
a
n l
a
n
gu
a
g
e
.
T
h
e
s
e
w
or
k
s
pr
ov
i
de
da
t
a
s
e
t
s
f
or
h
a
t
e
s
pe
e
c
h de
t
e
c
t
i
on
i
n
I
n
don
e
s
i
a
n
l
an
g
u
a
g
e f
r
o
m
T
w
i
t
t
er
.
T
h
es
e
w
o
r
k
s
al
s
o
u
s
ed
m
ac
h
i
n
e l
ea
r
n
i
n
g
ap
p
r
o
ach
t
o
t
ack
l
e t
h
i
s
p
r
o
b
l
em
.
B
as
i
cal
l
y
w
e
al
s
o
co
n
s
i
d
er
t
h
e
h
at
e
s
p
eech
d
et
ect
i
o
n
a
s
a
t
e
x
t
cl
a
s
s
i
f
i
cat
i
o
n
p
r
o
b
l
e
m
.
I
n
t
h
i
s
w
o
r
k
,
w
e
f
o
cu
s
o
n
t
h
e
p
r
o
b
l
em
o
f
cl
as
s
i
f
y
i
n
g
a t
w
eet
as
h
at
e s
p
eech
o
r
n
o
t
.
T
ex
t
cl
as
s
i
f
i
cat
i
o
n
t
ech
n
i
q
u
e
m
o
s
t
l
y
u
s
i
n
g
b
ag
o
f
w
o
r
d
s
f
eat
u
r
es
a
n
d
m
ac
h
i
n
e
l
ear
n
i
n
g
m
et
h
o
d
s
s
u
ch
as
N
aï
v
e B
ay
e
s
(
N
B
)
[
1
0
]
,
K
-
N
e
a
re
s
t
N
e
i
g
h
b
o
rs
(K
N
N
) [1
1
],
M
a
x
i
m
u
m
E
n
t
ro
p
y
(M
E
) [1
2
]
,
R
a
n
d
o
m
F
o
re
s
t
(R
F
) [1
3
],
o
r
S
u
p
p
o
rt
V
e
c
t
o
r
M
a
c
h
i
n
e
s
(S
V
M
) [1
] f
o
r
c
la
s
s
i
f
ic
a
tio
n
ta
s
k
.
I
n
t
hi
s
w
o
r
ks
,
w
e
us
e
d
e
n
s
e
m
b
l
e
m
e
t
h
od t
o t
a
c
k
l
e
t
hi
s
pr
obl
e
m
.
A
n
e
n
s
e
m
bl
e
of
c
l
a
s
si
f
i
e
r
s
i
s a
se
t
o
f
s
t
a
nd
-
a
l
o
ne
c
la
s
s
if
ie
r
s
w
h
ic
h
c
o
m
b
i
n
e
d
to
c
la
s
s
i
f
y
n
e
w
t
we
e
t
i
n or
de
r
t
o i
m
pr
ov
e
c
l
a
s
s
i
f
i
c
a
t
i
on
pe
r
f
or
m
a
n
c
e
[
1
4
]
.
I
n
g
en
er
al
t
e
x
t
cl
as
s
i
f
i
ca
t
i
o
n
,
s
ev
er
al
w
o
r
k
s
u
s
i
n
g
e
n
s
e
m
b
l
e
m
et
h
o
d
h
a
v
e b
een
co
n
d
u
ct
ed
an
d
r
ep
o
r
t
ed
th
a
t e
n
s
e
m
b
le
s
m
e
t
h
o
d
can
e
n
h
an
ce t
h
e cl
as
s
i
f
i
cat
i
o
n
p
er
f
o
r
m
an
ce (
e.
g
.
[
1
5
-
1
7
]).
S
ev
er
al
cl
as
s
i
f
i
er
t
h
a
t
b
een
us
e
d
i
n t
hi
s
e
ns
e
m
b
l
e
a
r
e
NB
,
KNN
,
M
E
,
RF
,
a
nd
SV
M
.
W
e ai
m
t
o
i
m
p
r
o
v
e
t
h
e p
er
f
o
r
m
a
n
ce o
f
s
o
m
e s
t
an
d
-
al
o
n
e cl
as
s
i
f
i
er
s
b
y
co
m
b
i
n
i
n
g
t
h
e
m
.
2.
R
ES
EA
R
C
H
M
ETH
O
D
A
s se
e
n
i
n
F
i
g
u
r
e 1
,
h
at
e
s
p
eech
d
et
ect
i
o
n
i
n
t
h
i
s
w
o
r
k
co
n
s
i
s
t
s
o
f
t
h
r
ee
m
ai
n
s
t
ag
es
:
1
)
p
r
e
p
r
oc
e
s
s
i
n
g;
2
)
t
r
a
i
ni
ng
s
o
m
e
s
t
a
nd
-
a
l
o
ne
cl
as
s
i
f
i
er
s
;
a
n
d
3
)
co
m
b
i
n
i
n
g
t
h
e cl
as
s
i
f
i
er
s
.
F
i
gu
r
e
1.
H
at
e S
p
eech
D
et
ect
i
o
n
F
l
o
w
ch
ar
t
2
.1
.
T
w
ee
t
P
rep
ro
ces
s
i
n
g
I
n
t
w
eet
p
r
ep
r
o
ces
s
i
n
g
,
t
h
er
e
ar
e s
o
m
e s
t
ep
s
t
o
b
e co
n
d
u
ct
ed
:
1
)
t
o
k
en
i
zat
i
o
n
;
2
)
f
i
l
t
er
i
n
g
;
3
)
s
te
m
m
i
n
g
; a
n
d
4
)
te
r
m
w
e
i
g
h
t
in
g
.
T
o
k
e
n
iz
a
tio
n
is
a
ta
s
k
o
f
s
p
litti
n
g
t
w
e
e
t
s
in
to
s
m
a
lle
r
u
n
its
c
a
lle
d
to
k
e
n
s
o
r
t
er
m
s
.
I
n
t
h
i
s
p
r
o
ces
s
,
cas
e f
o
l
d
i
n
g
a
n
d
cl
ean
s
i
n
g
ar
e al
s
o
co
n
d
u
ct
ed
.
C
a
s
e f
o
l
d
i
n
g
i
s
a p
r
o
ces
s
o
f
c
onv
e
r
t
i
ng
al
l
o
f
ch
ar
act
er
s
i
n
t
o
l
o
w
er
c
as
e.
I
n
t
h
e
cl
ea
n
s
i
n
g
p
r
o
ces
s
,
p
u
n
ct
u
at
i
o
n
,
n
u
m
b
er
s
,
h
t
m
l
t
ag
an
d
ch
ar
act
er
s
o
u
ts
id
e
o
f
th
e
a
lp
h
a
b
e
t
w
e
r
e
r
e
m
o
v
e
d
.
T
h
e
n
e
x
t
s
te
p
is
f
ilte
r
i
n
g
o
r
S
to
p
w
o
r
d
s
R
e
m
o
v
a
l.
S
to
p
w
or
ds
or
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
25
02
-
4752
I
nd
o
ne
s
i
a
n J
E
l
e
c
E
ng
&
C
o
m
p
Sc
i
,
Vo
l
.
11
, N
o
.
1
,
J
ul
y
201
8
:
2
64
–
299
296
u
n
i
nf
or
m
a
t
i
v
e
w
or
ds
w
e
r
e
r
e
m
o
v
e
d
i
n
t
h
i
s
s
t
e
p
ba
s
e
d
o
n
a
n
e
x
i
s
t
i
ng
s
t
opl
i
s
t
di
c
t
i
on
a
r
y.
I
n
t
h
i
s
w
or
k
,
w
e
s
t
opl
i
s
t
di
c
t
i
on
a
r
y
b
y
T
a
l
a
[
18
]
.
T
h
e
f
ou
r
t
h
s
t
e
p i
s
s
t
e
m
m
i
ng or
a
pr
oc
e
s
s
of
r
e
du
c
i
n
g
e
v
e
r
y
w
or
ds
t
o i
t
s
r
oot
or
ba
s
e
f
or
m
.
T
h
e
w
or
ds
‘
di
l
a
w
a
n’
,
‘
m
e
l
a
w
a
n
,
a
n
d ‘
pe
r
l
a
w
a
n
a
n w
i
l
l
be
c
onv
e
r
t
e
d
t
o
t
h
e
s
a
m
e
w
or
d ‘
l
a
w
a
n’
[
19]
-
[2
1
].
T
h
e
l
a
s
t
s
t
e
p i
n
pr
e
pr
oc
e
s
s
i
ng
i
s
w
or
d or
t
e
r
m
w
e
i
gh
t
i
n
g
.
I
n
t
h
i
s
w
or
k
,
w
e
u
s
e
ba
g
o
f
w
or
ds
(
B
O
W
)
f
eat
u
r
es
w
i
t
h
T
F
.
I
D
F
w
ei
g
h
t
i
n
g
.
T
F
.
I
D
F
i
s
t
h
e
m
o
s
t
p
o
p
u
l
ar
t
er
m
w
ei
g
h
t
i
n
g
m
e
th
o
d
i
n
te
x
t
c
la
s
s
if
ic
a
tio
n
[
2
2
]
.
T
F
.
I
D
F
i
s
a
c
o
m
b
i
na
t
i
o
n o
f
t
e
r
m
f
r
e
q
ue
nc
y
(
T
F
)
a
nd
i
nve
r
s
e
d
o
c
u
m
e
nt
f
r
e
q
ue
nc
y
(
I
D
F
)
.
T
he
T
F
.
I
D
F
w
e
i
g
h
t o
f
te
r
m
o
r
w
o
r
d
t in
t
w
e
e
t o
r
d
o
c
u
m
e
n
t d
is
c
a
lc
u
la
te
d
a
s
f
o
llo
w
s
:
(
)
)
l
og
1
(
)
l
og
1
(
)
,
(
,
+
⋅
+
=
⋅
t
d
d
t
df
N
f
d
t
I
DF
TF
w
h
er
e
d
t
f
,
i
s
t
h
e n
u
m
b
er
o
f
o
ccu
r
r
en
ces
o
f
t
er
m
t
i
n
t
w
eet
d
an
d
d
N
i
s
t
h
e
num
be
r
of
t
w
e
e
t
s
i
n c
or
pu
s
a
nd
t
df
i
s
t
h
e
n
u
m
b
er
o
f
t
w
ee
t
s
i
n
co
r
p
u
s
t
h
a
t
co
n
t
ai
n
s
t
er
m
t
.
F
i
n
a
l
l
y
,
t
h
i
s
s
t
a
g
e p
r
o
d
u
ce a b
ag
o
f
w
o
r
d
s
(
B
O
W
)
f
eat
u
r
es
w
h
i
ch
w
i
l
l
b
e u
s
ed
i
n
t
h
e n
e
x
t
s
t
ag
e.
2
.2
.
T
r
a
i
ni
ng
So
m
e
St
a
nd
-
a
l
o
n
e
C
la
s
s
if
ie
r
s
I
n
t
h
e
s
eco
n
d
s
t
ag
e,
s
e
v
er
al
p
o
p
u
l
ar
c
la
s
s
if
ie
r
s
is
tr
a
in
e
d
.
I
n
t
hi
s
w
o
r
k,
w
e
us
e
d
N
aï
v
e
B
ay
e
s
,
K
-
N
ear
es
t
N
ei
g
h
b
o
u
r
s
,
M
a
xi
m
u
m
E
nt
r
o
p
y
,
R
an
d
o
m
F
o
r
es
t
,
an
d
S
u
p
p
o
r
t
V
ect
o
r
Mach
i
n
es
.
F
o
r
N
aï
v
e B
ay
es
,
w
e
u
s
e
d
M
u
lti
n
o
m
ia
l
d
is
tr
ib
u
tio
n
a
s
it
p
r
o
v
e
s
to
s
h
o
w
g
o
o
d
p
e
r
f
o
r
m
a
n
c
e
in
te
x
t c
la
s
s
if
i
c
a
tio
n
.
Mea
n
w
h
i
l
e,
f
o
r
S
V
M
,
w
e
us
e
d
L
i
n
ear
k
er
n
el
f
o
r
t
h
e s
a
m
e r
eas
o
n
.
F
i
n
a
ll
y
,
th
e
c
la
s
s
i
f
ie
r
s
is
e
n
s
e
m
b
le
d
in
th
e
la
s
t s
ta
g
e
.
2
.3
.
C
o
m
b
in
in
g
t
h
e
C
la
s
s
if
ie
r
s
I
n t
he
l
a
st
s
t
ag
e,
s
e
v
er
al
cl
as
s
i
f
i
er
s
f
r
o
m
t
h
e p
r
ev
i
o
u
s
s
t
a
g
e i
s
co
m
b
i
n
ed
.
W
e co
n
d
u
c
t
ed
t
w
o
t
y
p
es
o
f
e
n
s
e
m
bl
e
m
e
t
h
ods
:
1
)
ha
r
d v
o
t
i
ng
;
a
n
d 2)
s
of
t
v
ot
i
ng
.
I
n h
ar
d
v
o
t
i
n
g
,
each
s
t
a
nd
-
a
l
o
ne
cl
a
s
s
i
f
i
er
h
a
s
o
n
e v
o
t
e.
A
s
s
een
i
n
F
i
g
u
r
e 2
,
t
h
e cat
eg
o
r
y
o
f
a
t
w
eet
is
s
el
ect
ed
b
y
m
aj
o
r
i
t
y
v
o
t
i
n
g
.
T
h
e cat
eg
o
r
y
s
el
ect
ed
i
s
t
h
e o
n
e
w
h
i
c
h ha
ve
a
m
a
j
o
r
i
t
y
,
t
ha
t
i
s
,
m
o
r
e
t
ha
n
ha
l
f
t
he
vo
t
e
s
.
M
ean
w
h
i
l
e i
n
s
o
f
t
v
o
t
i
n
g
,
av
er
ag
e
cat
e
g
o
r
y
p
r
o
b
a
b
ilitie
s
i
s
u
s
e
d
a
s
v
ot
i
ng
s
c
or
e
.
A
s
s
e
e
n
i
n
F
i
gur
e
3
,
t
he
f
i
na
l
c
a
t
e
go
r
y o
f
a
t
w
eet
i
s
t
he
o
ne
w
i
th
th
e
hi
g
he
s
t
v
ot
i
ng
s
c
or
e
or
av
er
ag
e p
r
o
b
a
b
i
l
i
t
y
f
r
o
m
eac
h
cl
as
s
i
f
i
er
s
.
F
i
gu
r
e
2.
H
a
r
d V
ot
i
n
g
E
ns
e
m
bl
e
M
et
h
o
d
F
i
gu
r
e
3.
S
o
f
t V
o
tin
g
E
n
s
e
m
b
le
M
e
th
o
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nd
o
ne
s
i
a
n J
E
l
e
c
E
ng
&
C
o
m
p
Sc
i
I
SSN
:
2502
-
4752
E
ns
e
m
bl
e
M
e
t
hod f
or
I
n
done
s
i
an T
w
i
t
t
e
r
H
at
e
Spe
e
c
h D
e
t
e
c
t
i
on
(
M
.
A
li F
a
u
z
i
)
297
3.
R
ES
U
LT
S
A
ND AN
AL
Y
S
I
S
W
e i
m
p
l
e
m
en
t
ed
t
h
e
e
x
p
er
i
m
en
t
s
u
s
i
n
g
S
ci
k
i
t
-
L
ear
n
[
2
3
]
.
W
e u
s
ed
T
w
i
t
t
er
h
at
e
s
p
eech
d
at
as
et
i
n
I
n
d
o
n
es
i
a
n
l
a
n
g
u
a
g
e t
h
at
h
av
e b
een
co
l
l
ect
ed
an
d
l
ab
el
l
ed
b
y
[
9
]
.
T
h
er
e ar
e 2
6
0
t
w
eet
s
l
ab
el
l
ed
as
h
at
e
s
p
eech
a
n
d
4
4
5
t
w
eet
s
l
ab
el
l
e
d
as
n
o
n
h
at
e
s
p
eech
.
W
e
k
ep
t
t
h
e
d
at
as
et
u
n
b
al
an
ced
i
n
t
h
e f
i
r
s
t
ex
p
er
i
me
n
t
.
F
o
r
t
h
e s
eco
n
d
e
x
p
er
i
m
e
n
t
,
w
e t
r
an
s
f
o
r
m
t
h
e
u
n
b
al
a
n
ced
d
at
as
et
i
n
t
o
a b
al
an
ced
o
n
e
u
s
i
n
g
an
u
n
d
er
s
a
m
p
l
i
n
g
m
et
h
o
d
.
W
e ch
o
o
s
e n
o
n
h
at
e
s
p
eech
t
w
eet
s
r
a
n
d
o
m
l
y
s
o
t
h
at
t
h
e
n
u
m
b
er
s
o
f
t
h
e
n
o
n
h
at
e s
p
eech
t
w
eet
s
b
eco
m
e t
h
e s
a
m
e n
u
m
b
er
o
f
t
h
e h
at
e
s
p
eech
t
w
eet
s
.
I
n t
he
e
xp
e
r
i
m
e
nt
s
,
w
e co
m
p
ar
ed
t
h
e r
es
u
l
t
s
o
f
s
t
an
d
-
al
o
n
e cl
as
s
i
f
i
er
s
w
i
t
h
o
ur
e
ns
e
m
b
l
e
m
e
t
ho
d
.
W
e
u
s
e
10 f
ol
d
c
r
os
s
v
a
l
i
da
t
i
on
,
w
hi
c
h
i
s
m
e
a
n
t
h
e
da
t
a
s
e
t
i
s
e
qu
a
l
l
y
di
v
i
de
d i
n
t
o 10 f
ol
ds
f
i
r
s
t
.
I
n
e
a
c
h
ite
r
a
tio
n
o
f
c
r
o
s
s
v
a
lid
a
tio
n
,
t
w
e
e
t
s
f
r
o
m
9
f
o
l
d
s
w
e
r
e
us
e
d
a
s
t
r
a
i
ni
ng d
a
t
a
a
nd
t
he
r
e
m
a
i
ni
ng
f
o
l
d
w
a
s
u
s
e
d
as
t
es
t
i
n
g
d
at
a.
W
e u
s
e av
er
ag
e F
1
M
eas
u
r
e as
t
h
e ev
al
u
at
i
o
n
m
et
h
o
d
i
n
t
h
i
s
ex
p
er
i
m
e
n
t
s
.
E
x
p
er
i
m
e
n
t
r
es
u
l
t
s
di
s
pl
a
y
e
d i
n
F
i
gu
r
e
4 a
n
d F
i
gur
e
5.
F
i
gu
r
e
4.
H
at
e S
p
eech
D
et
ect
i
o
n
P
e
r
f
o
r
m
a
nc
e
u
s
i
ng
U
n
b
al
an
ced
D
at
a
s
et
F
i
gu
r
e
5.
H
at
e S
p
eech
D
et
ect
i
o
n
P
er
f
o
r
m
a
n
ce u
s
i
n
g
B
al
an
ced
D
at
as
et
A
s
s
e
e
n i
n F
i
gu
r
e
4,
a
m
o
ng
a
l
l
s
t
a
n
d
-
a
l
o
n
e cl
a
s
s
i
f
i
er
s
,
N
B
h
as
t
h
e
b
es
t
p
er
f
o
r
m
an
ce
o
n
u
n
b
al
an
ced
d
at
as
et
co
m
p
ar
ed
w
i
t
h
o
t
h
er
s
t
an
d
-
al
o
n
e cl
as
s
i
f
i
er
b
y
7
8
.
3
%
F
1
m
ea
s
u
r
e.
S
V
M
p
er
f
o
r
m
ed
al
m
o
s
t
t
h
e s
a
m
e t
o
N
B
w
i
t
h F
1
m
e
a
s
u
r
e
78.
1%
.
I
t
i
s
c
l
e
a
r
t
o s
e
e
t
h
a
t
K
N
N
w
a
s
t
h
e
m
o
s
t
i
nf
e
r
i
or
c
l
a
s
s
i
f
i
e
r
w
i
t
h on
l
y
74.
2%
F
1
m
eas
u
r
e.
Mean
w
h
i
l
e,
R
F
an
d
M
E
p
er
f
o
r
m
ed
b
et
t
er
t
h
a
n
K
N
N
b
y
7
1
.
2
%
an
d
7
4
.
3
%
F
1
m
e
as
u
r
e r
es
p
ect
i
v
e
l
y.
A
l
m
o
s
t
al
l
o
f
t
h
e en
s
e
m
b
l
e
m
e
t
h
o
d
s
h
a
v
e h
i
g
h
er
F
1
m
eas
u
r
e o
v
er
s
t
an
d
-
al
o
n
e cl
as
s
i
f
i
er
s
o
n
u
n
b
al
a
n
ced
d
at
as
et
.
H
o
w
e
v
er
,
o
n
t
h
e h
ar
d
v
o
t
i
n
g
s
t
r
at
eg
y
w
i
t
h
5
cl
as
s
i
f
i
er
s
(
N
B
,
K
N
N
,
ME
,
R
F
,
S
V
M
)
,
w
h
o
s
e F
1
m
ea
s
u
r
e i
s
7
7
.
9
%
,
t
h
e en
s
e
m
b
l
e
m
et
h
o
d
s
ca
n
n
o
t
ex
cee
d
t
h
e
N
B
p
er
f
o
r
m
an
ce.
T
h
e d
eci
s
i
o
n
i
n
h
ar
d
v
o
tin
g
is
e
q
u
a
ll
y
d
e
te
r
m
i
n
e
d
b
y
a
l
l o
f
s
ta
n
d
-
a
l
o
ne
c
l
a
s
s
i
f
i
e
r
s
.
T
he
F
1
m
e
a
s
ur
e
o
f
ha
r
d
vo
t
i
ng
m
e
t
ho
d
us
ua
l
l
y
v
ar
i
es
b
et
w
ee
n
t
h
e
F
1
m
eas
u
r
e o
f
b
es
t
cl
as
s
i
f
i
er
a
n
d
t
h
e F
1
m
eas
u
r
e o
f
w
o
r
s
t
c
l
as
s
i
f
i
er
.
I
t
i
s
h
ar
d
f
o
r
t
he
ha
r
d
v
o
t
i
n
g
m
et
h
o
d
t
o
g
et
h
i
g
h
er
F
1
m
eas
u
r
e t
h
a
n
t
h
e b
es
t
cl
as
s
i
f
i
er
b
eacu
s
e t
h
e d
i
f
f
er
e
n
ce i
n
F
1
m
ea
s
u
r
e i
s
t
o
o
f
ar
b
et
w
een
t
h
e b
es
t
cl
a
s
s
i
f
i
er
(
7
8
.
3
%
)
an
d
t
h
e
w
o
r
s
t
cl
as
s
i
f
i
er
(
6
8
.
2
%
)
t
h
at
b
ee
n
c
o
m
b
i
n
ed
.
I
t
i
s
n
o
t
h
a
ppe
n
e
d
w
h
e
n
w
e
us
e
s
of
t
v
ot
i
ng
m
e
t
h
o
d
.
S
o
f
t
v
o
ti
n
g
m
e
t
h
o
d
w
it
h
5
c
la
s
s
i
f
ie
r
s
s
till s
u
r
p
a
s
s
t
h
e
p
e
r
f
o
r
m
a
n
c
e
o
f
a
ll s
ta
n
d
-
a
l
one
c
l
a
s
s
i
f
i
e
r
s
b
y
78.
9%
F
1
m
e
a
s
u
r
e
.
A
l
t
h
oug
h
c
o
m
bi
ni
ng
a
l
l
of
t
h
e
c
l
a
s
s
i
f
i
e
r
s
,
s
o
f
t
v
o
t
i
n
g
g
i
v
e v
o
t
e
s
f
o
r
each
cat
eg
o
r
y
b
as
ed
o
n
i
t
s
a
v
er
ag
e
p
r
o
b
ab
i
l
i
t
y
v
al
u
e
f
r
o
m
al
l
o
f
t
h
e cl
as
s
i
f
i
er
s
.
T
h
e
r
e
is
a
p
o
s
s
ib
ilit
y
t
h
a
t
w
in
n
in
g
c
a
te
g
o
r
ie
s
b
a
s
e
d
o
n
h
a
r
d
v
o
tin
g
w
ill lo
s
e
o
n
s
o
f
t v
o
ti
n
g
b
e
c
a
u
s
e
th
e
y
h
a
v
e
l
o
w
e
r
a
v
e
r
a
g
e
s
pr
oba
bi
l
i
t
y
t
ha
n
ot
h
e
r
c
a
t
e
g
or
y
.
S
of
t
v
ot
i
n
g s
i
m
pl
y
pr
ov
i
de
s
a
m
or
e
r
obus
t
v
ot
i
ng
s
c
h
e
m
e
a
s
it is
o
f
te
n
r
e
d
u
c
es
o
v
er
f
i
t
a
n
d
cr
eat
es
a s
m
o
o
t
h
er
m
o
d
el
.
T
h
e en
s
e
m
b
l
e
m
e
t
h
o
d
s
b
y
u
s
i
n
g
o
n
l
y
t
h
r
ee b
es
t
cl
as
s
i
f
i
e
r
s
(
N
B
,
S
V
M
,
an
d
R
F
)
h
av
e t
h
e b
es
t
p
e
r
f
o
r
m
a
nc
e
w
he
n u
s
i
ng
ha
r
d
vo
t
i
n
g o
r
s
o
f
t
vo
t
i
n
g.
H
a
r
d
vo
t
i
ng a
nd
s
o
f
t
vo
t
i
ng b
a
s
e
d
o
n t
hi
s
s
c
he
m
e
ha
ve
t
h
e
s
a
m
e
F
1
m
eas
u
r
e,
7
9
.
8
%
.
S
i
n
ce en
s
e
m
b
l
e
m
et
h
o
d
i
s
af
f
ect
ed
b
y
t
h
e
cl
as
s
i
f
i
er
s
t
h
at
co
m
p
i
l
ed
i
t
,
u
s
i
n
g
o
n
l
y
t
h
e b
es
t
cl
as
s
i
f
i
er
can
i
m
p
r
o
v
e t
h
e p
o
s
s
i
b
i
l
i
t
y
o
f
en
s
e
m
b
l
e m
e
t
h
o
d
t
o
g
et
b
et
t
er
p
er
f
o
r
m
an
ce.
M
ean
w
h
i
l
e,
t
h
e
r
es
u
l
t
o
f
t
h
e
s
eco
n
d
ex
p
er
i
m
en
t
,
w
h
i
c
h
i
s
u
s
i
n
g
b
al
an
ced
d
at
a
s
et
,
ca
n
b
e s
een
i
n
F
i
g
u
r
e 5
.
A
s
p
r
ed
i
ct
ed
,
al
l
o
f
t
h
e cl
a
s
s
i
f
i
cat
i
o
n
m
et
h
o
d
g
o
t
h
i
g
h
er
F
1
m
eas
u
r
e o
n
b
al
a
n
c
ed
d
at
as
et
.
K
N
N
i
s
s
t
i
l
l
t
h
e
w
or
s
t
c
l
a
s
s
i
f
i
e
r
w
i
t
h F
1
m
e
a
s
u
r
e
76.
8%
.
R
F
pe
r
f
o
m
e
d
s
l
i
g
h
t
l
y
be
t
t
e
r
w
i
t
h 77.
6%
F
1
m
e
a
s
u
r
e
.
S
ur
p
r
i
s
i
ngl
y,
M
E
ha
s
t
h
e
b
es
t
p
er
f
o
r
m
a
n
ce
w
i
t
h
8
4
.
1
%
F
1
m
eas
u
r
e
v
al
u
e.
N
B
a
n
d
S
V
M
ar
e s
t
i
l
l
b
el
o
w
M
E
w
i
t
h
o
n
l
y
s
l
i
g
h
t
d
i
f
f
er
en
ce.
A
l
l
o
f
t
h
e
en
s
e
m
b
l
e
m
et
h
o
d
h
av
e
al
m
o
s
t
t
h
e
s
a
m
e
F
1
m
e
as
u
r
e
v
a
l
u
e
a
n
d
al
s
o
p
er
f
o
r
m
b
et
t
er
t
h
an
al
m
o
s
t
al
l
o
f
t
h
e
s
t
a
n
d
-
al
o
n
e
cl
as
s
i
f
i
er
s
.
T
h
es
e t
w
o
e
x
ep
er
i
m
e
n
t
s
s
h
o
w
ed
t
h
at
w
e
ca
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
25
02
-
4752
I
nd
o
ne
s
i
a
n J
E
l
e
c
E
ng
&
C
o
m
p
Sc
i
,
Vo
l
.
11
, N
o
.
1
,
J
ul
y
201
8
:
2
64
–
299
298
i
m
p
r
o
ve
t
he
p
e
r
f
o
r
m
a
nc
e
b
y
us
i
n
g e
ns
e
m
b
l
e
m
e
t
ho
d
e
ve
n
i
f
no
t
s
i
g
ni
f
i
c
a
nt
.
N
e
ve
r
t
he
l
e
s
s
,
e
n
s
e
m
b
l
e
m
e
t
ho
d
s
u
r
el
y
r
ed
u
ce t
h
e j
eo
p
ar
d
y
o
f
ch
o
o
s
i
n
g
a
w
eek
cl
a
s
s
i
f
i
er
t
o
b
e u
s
ed
f
o
r
d
et
ect
i
n
g
n
e
w
t
w
e
et
s
.
4.
CO
NCL
U
S
I
O
N
I
n
t
h
i
s
s
t
ud
y,
w
e
we
u
s
e
d
en
s
e
m
b
l
e m
et
h
o
d
t
o
f
o
r
H
at
e S
p
eech
D
et
ect
i
o
n
i
n
I
n
d
o
n
e
s
i
a
n
l
an
g
u
a
g
e.
W
e em
p
l
o
y
ed
f
i
v
e s
t
a
n
d
-
a
lo
n
e
c
la
s
s
i
f
ic
a
tio
n
a
lg
o
r
it
h
m
s
,
in
c
l
u
d
in
g
N
a
ïv
e
B
a
y
e
s
,
K
-
N
ear
es
t
N
ei
g
h
b
o
u
r
s
,
M
ax
i
m
u
m
E
n
t
r
o
p
y
,
R
an
d
o
m
F
o
r
es
t
,
an
d
S
u
p
p
o
r
t
V
ect
o
r
Mach
i
n
es
,
a
n
d t
w
o e
ns
e
m
bl
e
m
e
t
h
ods
on
T
w
i
t
t
e
r
h
at
e s
p
eech
d
at
as
e
t
.
B
y
u
s
i
n
g
u
n
b
al
an
ce d
at
as
et
,
t
h
e ex
p
er
i
m
e
n
t
r
es
u
l
t
s
s
h
o
w
t
h
at
N
aï
v
e B
a
y
e
s
o
f
f
er
ed
t
h
e b
es
t
p
er
f
o
r
m
a
n
ce a
m
o
n
g
al
l
f
i
v
e s
t
an
d
-
al
o
n
e cl
as
s
i
f
i
er
s
w
i
t
h
F
1
m
ea
s
u
r
e v
a
l
u
e 7
8
.
3
%
.
T
h
e ex
p
er
i
m
e
n
t
al
r
es
u
l
t
s
al
s
o
s
h
o
w
t
h
at
e
n
s
e
m
b
l
e t
ec
h
n
i
q
u
e ca
n
i
m
p
r
o
v
e
t
h
e
cl
as
s
i
f
i
cat
i
o
n
p
er
f
o
r
m
an
ce.
T
h
e b
es
t
r
es
u
l
t
i
s
ach
i
ev
ed
w
h
e
n
u
s
i
n
g
e
n
s
e
m
b
l
e o
f
t
h
r
ee b
es
t
cl
as
s
i
f
i
er
(
N
aï
v
e B
a
y
es
,
S
u
p
p
o
r
t
V
ect
o
r
M
ach
i
n
e,
a
n
d
R
an
d
o
m
F
o
r
es
t
)
w
i
t
h
F
1
m
e
a
s
u
r
e
79.
8%
.
M
ean
w
h
i
l
e,
as
p
r
ed
i
ct
ed
,
al
l
o
f
t
h
e cl
a
s
s
i
f
i
cat
i
o
n
m
et
h
o
d
g
o
t
h
i
g
h
er
F
1
m
eas
u
r
e
w
h
en
us
i
ng
b
al
an
ced
d
at
as
et
.
S
u
r
p
r
i
s
i
n
g
l
y
,
Max
i
m
u
m
E
n
t
r
o
p
y
h
as
t
h
e
b
es
t
p
er
f
o
r
m
a
n
ce i
n
t
h
i
s
s
eco
n
d
ex
p
er
i
m
e
n
t
w
i
t
h
8
4
.
1
%
F
1
m
eas
u
r
e v
al
u
e.
U
s
i
n
g
b
al
a
n
ced
d
at
as
et
,
al
l
o
f
t
h
e en
s
e
m
b
l
e
m
et
h
o
d
h
av
e al
m
o
s
t
t
h
e s
a
m
e F
1
m
eas
u
r
e
v
al
u
e
an
d
al
s
o
p
er
f
o
r
m
b
et
t
er
t
h
a
n
al
m
o
s
t a
ll o
f
t
h
e
s
ta
n
d
-
al
o
n
e
cl
as
s
i
f
i
er
s
.
T
h
es
e
t
w
o
ex
ep
er
i
m
en
t
s
s
ho
w
e
d
t
ha
t
u
s
i
ng e
ns
e
m
b
l
e
m
e
t
ho
d
c
a
n i
m
p
r
o
ve
t
he
p
e
r
f
o
r
m
a
nc
e
o
f
t
he
s
ys
t
e
m
.
A
l
t
ho
u
gh
t
he
i
m
p
r
o
ve
m
e
n
t
i
s
n
o
t
s
i
g
n
i
f
i
ca
n
t
,
u
s
i
n
g
en
s
e
m
b
l
e
m
et
h
o
d
can
r
ed
u
ce t
h
e
r
i
s
k
o
f
s
el
ect
i
n
g
a
p
o
o
r
cl
as
s
i
f
i
er
t
o be
u
s
e
d
f
or
d
et
ect
i
n
g
n
e
w
t
w
eet
s
as
h
at
e s
p
eech
o
r
n
o
t
.
I
n t
he
f
ut
ur
e
w
o
r
k,
i
ns
t
e
a
d
o
f
o
nl
y
us
i
ng B
O
W
f
e
a
t
ur
e
s
,
a
p
p
l
yi
n
g e
n
s
e
m
b
l
e
s
o
f
f
e
a
t
ur
e
s
e
t
m
a
y
b
e
a
p
r
o
m
i
s
i
n
g
d
i
r
ect
i
o
n
t
o
g
et
b
et
t
er
p
er
f
o
r
m
an
ce.
S
o
m
e f
eat
u
r
e s
et
s
u
c
h
as
n
-
gr
a
m
,
l
e
x
i
c
o
n,
P
O
S
t
a
g
gi
n
g,
t
e
x
ua
l
f
eat
u
r
e o
r
t
w
i
t
t
er
s
p
eci
f
i
c
f
eat
u
r
es
ca
n
b
e ap
p
l
i
ed
f
o
r
i
m
p
r
o
v
e
m
e
n
t
.
A
n
o
t
h
er
t
y
p
e
s
o
f
f
eat
u
r
e l
i
k
e W
o
r
d
2
V
ec
o
r
P
a
r
ag
r
ap
h
2
V
ec al
s
o
can
b
e ap
p
l
i
ed
i
n
t
h
e f
u
t
u
r
e.
R
EF
ER
EN
C
ES
[
1]
W
a
r
n
e
r
W
,
H
i
r
s
ch
b
er
g
J
,
“
D
et
ect
i
n
g
h
at
e s
p
eech
o
n
t
h
e
W
o
r
l
d
W
i
d
e
W
e
b
”
.
I
n
P
r
oc
e
e
di
ngs
of
t
he
Se
c
on
d
W
or
k
s
hop on L
a
ng
ua
ge
i
n Soc
i
al
M
e
di
a
20
12
J
un
7 (
pp
.
1
9
-
26)
.
A
s
s
oc
i
a
t
i
on f
or
C
om
put
a
t
i
o
na
l
L
i
ng
ui
s
t
i
c
s
.
[
2]
S
ito
r
u
s
A
P
,
M
ur
f
i
H
,
N
ur
r
ohm
a
h S
,
A
k
ba
r
A
,
“
S
en
s
i
n
g
T
r
en
d
i
n
g
T
o
p
i
cs
i
n
T
w
i
t
t
er
f
o
r
G
r
eat
er
J
ak
ar
t
a
A
r
ea
”,
I
nt
e
r
nat
i
o
nal
J
our
n
al
of
E
l
e
c
t
r
i
c
al
an
d C
om
pu
t
e
r
E
n
gi
ne
e
r
i
n
g (
I
J
E
C
E
)
.
20
17
F
e
b 1
;
7(
1)
:
3
30
-
6.
[
3]
B
u
r
n
a
p
P
,
W
illia
m
s
M
L
,
“
C
y
b
er
h
at
e s
p
eech
o
n
t
w
i
t
t
er
:
A
n
ap
p
l
i
cat
i
o
n
o
f
m
ach
i
n
e cl
as
s
i
f
i
cat
i
o
n
an
d
s
t
at
i
s
t
i
ca
l
m
ode
l
i
ng
f
or
pol
i
c
y
a
nd de
c
i
s
i
on
m
a
k
i
ng
”
.
P
o
l
i
cy
&
I
n
t
er
n
et
.
201
5
J
un 1;
7(
2)
:
22
3
-
42
.
[
4]
B
ad
j
at
i
y
a
P
,
G
u
p
t
a S
,
G
u
p
t
a M
,
V
ar
m
a V
,
“
D
e
ep
l
ear
n
i
n
g
f
o
r
h
at
e s
p
eech
d
et
ect
i
o
n
i
n
t
w
e
et
s
”
.
In
P
r
oc
e
e
di
ngs
o
f
t
he
2
6t
h I
nt
e
r
nat
i
on
al
C
o
nf
e
r
e
nc
e
on
W
or
l
d
W
i
de
W
e
b C
om
p
ani
o
n
,
2
01
7 A
p
r
3 (
pp
.
7
59
-
7
60)
.
I
nt
e
r
na
t
i
o
na
l
W
or
l
d
W
i
de
W
e
b C
onf
e
r
en
ces
S
t
eer
i
n
g
C
o
m
m
i
t
t
ee.
[
5]
W
as
ee
m
Z
,
H
o
v
y
D
,
“
H
at
e
f
u
l
S
y
m
b
o
l
s
o
r
H
at
ef
u
l
P
eo
p
l
e?
P
r
e
d
i
ct
i
v
e F
eat
u
r
es
f
o
r
H
at
e S
p
eech
D
et
ect
i
o
n
o
n
Tw
itte
r
”
.
In
S
RW
@
H
L
T
-
N
A
A
C
L
2016
J
un
1
2 (
p
p.
88
-
9
3)
.
[
6]
K
w
o
k I
,
W
a
n
g Y
,
L
o
cat
e t
h
e H
at
e:
D
et
ect
i
n
g
T
w
e
et
s
ag
ai
n
s
t
B
l
ack
s
.
I
n
A
A
A
I
2
01
3
J
u
l
1
4.
[
7]
B
ar
b
o
s
a L
,
F
en
g
J
,
“
R
obus
t
s
e
nt
i
m
e
nt
de
t
e
c
t
i
on
on t
w
i
t
t
e
r
f
r
o
m
bi
a
s
e
d a
nd no
i
s
y
da
t
a
”
.
In
P
r
oc
e
e
di
ngs
of
t
he
2
3r
d
I
nt
e
r
nat
i
o
nal
C
o
nf
e
r
e
nc
e
on C
om
put
at
i
on
al
L
i
ng
ui
s
t
i
c
s
:
P
os
t
e
r
s
201
0 A
ug
23 (
pp.
36
-
44)
.
A
s
s
oc
i
a
t
i
on
f
or
C
o
m
p
u
ta
tio
n
a
l L
in
g
u
is
t
ic
s
.
[
8]
S
.
H
.
P
r
a
tiw
i,
“D
et
ect
i
o
n
o
f
H
at
e S
p
eech
ag
ai
n
s
t
R
el
i
g
i
o
n
o
n
T
w
ee
t
i
n
t
h
e I
n
d
o
n
es
i
an
L
an
g
u
ag
e U
s
i
n
g
N
aï
v
e B
a
y
es
A
l
g
or
i
t
hm
a
nd S
u
pp
or
t
V
e
c
t
or
M
a
c
hi
ne
”
,
B
.
S
c.
T
es
i
s
,
U
n
i
v
er
s
i
t
as
I
n
d
o
n
es
i
a,
I
n
d
o
n
es
i
a,
2
0
1
6
.
[
9]
A
l
f
i
n
a I
,
M
u
l
i
a R
,
F
an
an
y
M
I
,
E
k
an
at
a Y
.
“
H
at
e S
p
ee
ch
D
et
ect
i
o
n
i
n
t
h
e I
n
d
o
n
es
i
an
L
an
g
u
ag
e:
A
D
at
a
s
et
an
d
P
r
e
lim
in
a
r
y
S
tu
d
y
”
.
In
A
dv
anc
e
d C
om
pu
t
e
r
Sc
i
e
nc
e
and I
nf
or
m
at
i
o
n Sy
s
t
e
m
s
(
I
C
A
C
SI
S)
,
20
1
7 I
nt
e
r
nat
i
o
na
l
C
onf
e
r
e
nc
e
o
n 2
01
7
. I
E
E
E
.
[
1
0]
F
a
u
z
i
M
A,
Ar
i
f
i
n
AZ
,
Go
s
a
r
i
a
S
C,
“
I
n
d
o
n
e
si
a
n
Ne
w
s C
l
a
ssi
f
i
c
a
t
i
o
n
Usi
n
g
N
aï
v
e B
a
y
es
an
d
T
w
o
-
P
h
as
e F
eat
u
r
e
S
el
ect
i
o
n
M
o
d
el
”
.
I
n
do
ne
s
i
an J
o
ur
nal
of
E
l
e
c
t
r
i
c
al
E
ngi
ne
e
r
i
ng
a
nd C
om
p
ut
e
r
Sc
i
e
nc
e
.
2
01
7 D
e
c
1;
8(
3)
.
[
1
1]
S
uha
r
no C
F
,
F
a
uz
i
M
A
,
P
e
r
da
n
a
R
S
,
“
K
l
as
i
f
i
k
a
s
i
T
ek
s
B
ah
as
a
I
n
d
o
n
es
i
a P
ad
a D
o
k
u
m
en
P
en
g
ad
u
an
S
am
b
at
O
nl
i
ne
M
e
ng
g
una
k
a
n
M
e
t
od
e
K
-
N
ear
es
t
N
ei
g
h
b
o
r
s
D
an
C
h
i
-
S
qua
r
e
”
.
S
ys
t
em
i
c:
I
n
f
o
r
m
at
i
on
S
y
s
t
e
m
and
I
nf
or
m
at
i
c
s
J
our
n
al
.
2
01
7 D
e
c
7;
3(
1)
:
25
-
32.
[
1
2]
E
l
-
H
al
ees
A
M
,
“
A
r
ab
i
c t
ex
t
cl
as
s
i
f
i
cat
i
o
n
u
s
i
n
g
m
ax
i
m
u
m
en
t
r
o
p
y
”
.
I
U
G
J
our
nal
of
N
a
t
ur
a
l
St
udi
e
s
.
20
15
D
e
c
5;
15(
1)
.
[
1
3]
W
u
Q,
Ye
Y,
Z
h
a
n
g
H,
Ng
M
K
,
Ho
S
S
,
“
F
o
r
es
T
ex
t
er
:
an
ef
f
i
ci
en
t
r
an
d
o
m
f
o
r
es
t
al
g
o
r
i
t
h
m
f
o
r
i
m
b
al
an
ced
t
ex
t
cat
eg
o
r
i
zat
i
o
n
”
.
K
now
l
e
dge
-
B
a
s
ed
S
ys
t
em
s
.
2
01
4 S
e
p 3
0;
6
7:
10
5
-
16
.
[
1
4]
R
o
li F
.
“
M
u
l
tip
le
c
la
s
s
if
ie
r
s
y
s
te
m
s
”,
E
nc
y
c
l
ope
di
a of
B
i
om
e
t
r
i
c
s
.
20
15:
11
42
-
7.
[
1
5]
Ad
e
v
a
J
G,
At
x
a
J
P
,
Ca
rri
l
l
o M
U
,
Z
e
ng
ot
i
t
a
be
ng
oa
E
A
,
“
A
u
to
m
a
tic
te
x
t c
la
s
s
i
f
ic
a
tio
n
to
s
u
p
p
o
r
t s
y
s
t
e
m
a
tic
r
e
v
ie
w
s
in
m
e
d
ic
in
e
”,
E
x
pe
r
t
S
y
s
t
e
m
s
w
i
t
h A
ppl
i
c
a
t
i
o
ns
.
2
01
4 M
a
r
31
;
4
1(
4
)
:
14
98
-
50
8.
[
1
6]
Do
n
g
YS
,
Ha
n
KS
,
“
A
co
m
p
ar
i
s
o
n
o
f
s
ev
er
al
en
s
e
m
b
l
e
m
et
h
o
d
s
f
o
r
t
ex
t
cat
e
g
o
r
i
zat
i
o
n
”
.
In
S
er
vi
c
e
s
C
om
put
i
ng,
200
4.
(
SC
C
2
00
4)
.
P
r
oc
e
e
di
n
gs
.
2
004
I
E
E
E
I
nt
e
r
n
at
i
on
al
C
o
nf
e
r
e
n
c
e
on
,
20
04 S
e
p 15
(
pp
.
4
19
-
4
2
2
)
.
I
E
E
E
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nd
o
ne
s
i
a
n J
E
l
e
c
E
ng
&
C
o
m
p
Sc
i
I
SSN
:
2502
-
4752
E
ns
e
m
bl
e
M
e
t
hod f
or
I
n
done
s
i
an T
w
i
t
t
e
r
H
at
e
Spe
e
c
h D
e
t
e
c
t
i
on
(
M
.
A
li F
a
u
z
i
)
299
[
1
7]
La
r
k
e
y
LS
,
C
r
o
f
t
WB
,
“
C
o
m
b
in
in
g
c
la
s
s
if
ie
r
s
in
te
x
t c
a
te
g
o
r
iz
a
tio
n
”
.
In
P
r
oc
e
e
di
ngs
of
t
he
19t
h
an
nu
al
i
nt
e
r
nat
i
o
nal
A
C
M
SI
G
I
R
c
onf
e
r
e
nc
e
on R
e
s
e
ar
c
h
an
d
de
v
e
l
opm
e
nt
i
n i
nf
or
m
a
t
i
o
n r
e
t
r
i
e
v
al
,
19
96
A
u
g 18
(
pp.
2
89
-
2
9
7
).
A
CM
.
[
1
8]
T
al
a F
,
“
A
s
t
ud
y
of
s
t
e
m
m
i
ng
e
f
f
e
c
t
s
on i
nf
or
m
a
t
i
on r
e
t
r
i
e
v
a
l
i
n B
a
ha
s
a
I
ndo
ne
s
i
a
”
.
I
ns
t
i
t
ut
e
f
or
L
o
gi
c
,
L
an
gu
age
and
C
om
put
at
i
on,
U
n
i
v
er
s
i
t
ei
t
v
an
A
m
s
t
er
d
a
m
,
T
h
e N
et
h
er
l
a
nds
.
2
003
J
ul
.
[
1
9]
P
r
a
m
uk
a
nt
or
o E
S
,
F
a
uz
i
M
A
,
“
C
o
m
p
a
r
a
tiv
e
a
n
a
l
y
s
is
o
f
s
tr
in
g
s
i
m
ila
r
ity
a
n
d
c
o
r
p
u
s
-
b
a
s
e
d
s
im
ila
r
it
y
f
o
r
a
u
to
m
a
tic
e
ssa
y
s
c
o
r
i
n
g
sy
st
e
m
o
n
e
-
l
ear
n
i
n
g
g
am
i
f
i
cat
i
o
n
”
,
In
A
dv
a
nc
e
d C
om
put
e
r
Sc
i
e
nc
e
a
nd
I
nf
or
m
at
i
o
n
Sy
s
t
e
m
s
(
ICA
CS
IS
),
2
0
1
6
In
t
e
r
n
a
t
i
on
al
C
onf
e
r
e
nc
e
o
n
,
20
16 O
c
t
15 (
pp
.
1
49
-
1
5
5
)
. I
E
E
E
.
[
2
0]
F
a
u
z
i M
A
,
A
r
if
in
A
,
Y
u
n
ia
r
ti A
,
“
T
e
r
m
W
e
i
g
ht
i
ng
B
e
r
ba
s
i
s
I
n
de
k
s
B
uk
u
da
n K
e
l
a
s
unt
uk
P
e
r
a
ng
k
i
ng
a
n D
ok
um
e
n
B
e
r
ba
ha
s
a
A
r
a
b.
L
ont
a
r
K
om
put
e
r
”,
J
ur
nal
I
l
m
i
a
h T
e
k
n
ol
o
gi
I
nf
or
m
as
i
.
2
01
3;
5(
2)
.
[
2
1]
F
a
u
z
i
, M
.A
., U
t
o
m
o
, D
.C
., S
e
t
i
a
w
a
n,
B
.
D
.
a
nd P
r
a
m
uk
a
nt
or
o,
E
.
S
,
“
A
u
to
m
a
tic
Es
s
a
y
S
c
o
r
in
g
S
y
s
te
m
U
s
in
g
N
-
G
r
a
m
an
d
C
o
s
i
n
e
S
i
m
i
l
ar
i
t
y
f
o
r
G
a
m
i
f
i
cat
i
o
n
B
as
ed
E
-
L
ear
n
i
n
g
”
,
In
P
r
oc
e
e
di
ngs
of
t
he
I
nt
e
r
na
t
i
o
nal
C
on
f
e
r
e
nc
e
on A
dv
a
nc
e
s
i
n I
m
a
ge
P
r
oc
e
s
s
i
n
g
(
p
p.
15
1
-
1
5
5
).
A
CM
.
[
2
2]
F
au
zi
M
A,
Ar
i
f
i
n
AZ
,
Y
u
n
i
a
r
t
i
A,
“
A
r
a
bi
c
B
ook
R
e
t
r
i
e
v
a
l
us
i
ng C
l
a
s
s
a
nd B
o
ok
I
nde
x
B
a
s
e
d T
e
r
m
W
e
i
g
ht
i
ng
”,
I
nt
e
r
nat
i
o
nal
J
our
n
al
of
E
l
e
c
t
r
i
c
al
an
d C
om
pu
t
e
r
E
n
gi
ne
e
r
i
n
g (
I
J
E
C
E
)
,
2
01
7 D
e
c
1;
7(
6)
.
[
2
3]
P
ed
r
eg
o
s
a F
,
V
ar
o
q
u
au
x
G
,
G
r
a
m
f
o
r
t
A
,
M
i
ch
el
V
,
T
h
i
r
i
o
n
B
,
G
r
i
s
el
O
,
B
l
o
n
d
el
M
,
P
r
et
t
en
h
o
f
er
P
,
W
ei
s
s
R
,
D
ubo
ur
g
V
,
V
a
nde
r
pl
a
s
J
,
“
S
c
ik
it
-
l
ear
n
:
M
ach
i
n
e l
ear
n
i
n
g
i
n
P
y
t
h
o
n
”
,
J
our
n
al
of
M
ac
h
i
ne
L
e
ar
ni
n
g R
e
s
e
ar
c
h
.
201
1;
12(
O
c
t
)
:
28
25
-
3
0.
Evaluation Warning : The document was created with Spire.PDF for Python.